Vés al contingut (premeu Retorn)

Programació del Seminari: Any 2012

  • Dimecres 5 de desembre de 2012, Hora 12:30
Shrinkage, Pretest, and Penalty Estimation Strategies in Generalized Linear Models
Syed Ejaz Ahmed, Department of Mathematics, Brock University, Canada.

 

  • Dimarts 20 de novembre de 2012, Hora: 12:30
Complementarity Problems and Variational Inequalities
Joaquim J. Júdice, Instituto de Telecomunicações, Portugal

  • Divendres 16 de novembre de 2012, Hora: 12:30
Routing problems with profits
M. Grazia Speranza. University of Brescia, Italy

  • Dijous 15 de novembre de 2012, Hora: 12:00
Recent advances in phase II/III clinical trials: Study Design and Methods of Analysis
Sue Todd, Department of Mathematics and Statistics, University of Reading, UK

  • Divendres, 19 d'octubre de 2012, Hora: 12:30
Un algoritmo de branch-and-cut para un problema de rutas con un camión, un remolque, y puntos de recarga
Enrique Benavent, Departament d'Estadística i Investigació Operativa, Universitat de València

  • Divendres, 21 de setembre de 2012, Hora: 12:30
Generating solutions to practical decision problems by dynamic programming
Eligius M.T. Hendrix, Universitat de Malaga/Wageningen University

 

  • Dijous, 5 de juliol de 2012, Hora: 13:00
Functional Analysis of Cardiovascular Data
Emilia Bagiella, Department of Biostatistics, Mailman School of Public Health, Columbia University, New York.

 

  • Divendres, 29 de juny de 2012, Hora: 12:30
Spatio-temporal modelling of wildfires in Catalonia through log Gaussian Cox processes
Marc Saez, Research Group on Statistics, Applied Economics and Health (GRECS), University of Girona, Spain. CIBER of Epidemiology and Public Health (CIBERESP).

  • Divendres, 25 de maig de 2012, Hora: 12:30
Data Analytics of Narrative: Pattern Recognition in Text, and Text Synthesis, Supported by the Correspondence Analysis Platform.
Fionn Murtagh, Department of Computer Science, Royal Holloway, University of London

  • Divendres, 18 de maig de 2012, Hora: 12:30
Locating Lottery Retail Stores in Spain
Daniel Serra, Departament d'Economia i Empresa, Universitat Pompeu Fabra

  • Divendres, 4 de maig de 2012, Hora: 12:30
Statistical Aspects of Google Flu Trends
Samantha Cook, Departament d'Economia i Empresa, Universitat Pompeu Fabra

  • Divendres, 27 d'abril de 2012, Hora: 12:30
Detection and investigation of carcinogenic risk exposures using the observed temporal pattern of the associated events
Rina Chen, The Israel Institute for Biological Research, Ness-Ziona, Israel

  • Divendres, 20 d'abril de 2012, Hora: 12:30
Análisis forense de datos electorales
Raúl Jiménez, Departamento de Estadística, Universidad Carlos III de Madrid

  • Dimecres, 18 d'abril de 2012, Hora: 12:30
Stochastic programming with mixed-integer variables
Maarten van der Vlerk, Department of Operations, University of Groningen, Holanda

  • Dimarts, 17 d'abril de 2012, Hora: 12:30
Spatial prediction of nonstationary functional processes
Ramón Giraldo, Departamento de Estadística, Universidad Nacional de Colombia

  • Dijous, 15 de març de 2012, Hora: 12:30
Global optimization strategies for detailed kinetic models in systems biology
Albert Sorribas, Department de Ciències Mèdiques Bàsiques, Universitat de Lleida

  • Divendres, 27 de gener de 2012, Hora: 12:30
El problema del cartero chino con beneficio máximo
Ángel Corberán, Dpto. de Estadística e Investigación Operativa, Universidad de Valencia.




Shrinkage, Pretest, and Penalty Estimation Strategies in Generalized Linear Models
CONVIDAT:
Syed Ejaz Ahmed

IDIOMA: Anglès

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: Dimecres, 5 de desembre de 2012. Hora: 12:30
RESUM:
I will discuss estimation strategies in generalized linear models when there are many potential predictors and some of them may not have influence on the response of interest. In the context of two competing models where one model includes all predictors and the other restricts variable coefficients to a candidate linear subspace based on subject matter or prior knowledge, we investigate the relative performances of Stein type shrinkage, pretest, and penalty estimators (L1GLM, adaptive L1GLM, and SCAD) with respect to the unrestricted maximum likelihood estimator (MLE). The asymptotic properties of the pretest and shrinkage estimators including the derivation of asymptotic distributional biases and risks are established. In particular, we give conditions under which the shrinkage estimators are asymptotically more efficient than the unrestricted MLE. A Monte Carlo simulation study show that the mean squared error (MSE) of an adaptive shrinkage estimator is comparable to the MSE of the penalty estimators in many situations and in particular performs better than the penalty estimators when the dimension of the restricted parameter space is large. The Steinian shrinkage and penalty estimators all improve substantially on the unrestricted MLE. A real data set analysis is also presented to compare the suggested methods.

El ponent: clicar aquí per més informació sobre les linies d'investigació d'en Ejaz Ahmed



Inici

Complementarity Problems and Variational Inequalities
CONVIDAT:
Joaquim J. Júdice
IDIOMA: Anglès
LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)
DATA: Dimarts, 20 de novembre de 2012. Hora: 12:30
RESUM:
A Complementarity Problem (CP) is an optimization problem containing a number of comple-
mentary variables xi and wi satisfying constraints of the form xiwi = 0. This problem has found
many applications in different areas of science, engineering, economics and finance. A number
of extensions of this problem have been considered in the literature, namely Variational Inequali-
ties (VI), Optimization Problems with Complementarity Constraints (OPTCC) and more recently
Eigenvalue Complementarity Problems (EiCP). In this talk all these problems are defined together
with a brief summary of their applications. Some existence results for these problems are reviewed.
A number of efficient direct and iterative algorithms have been proposed to process the CP, VI,
OPTCC and EiCP by finding stationary points of appropriate merit functions. These algorithms
are introduced, together with their benefits and drawbacks. In particular, these techniques require
special classes of matrices for guaranteeing that those stationary points are solutions of the prob-
lems. In general, only enumerative algorithms are able to solve CP, VI, OPTCC and EiCP. The
most important of these methods are also presented. A number of numerical results are reported
to illustrate the efficacy and efficiency of the techniques discussed in the talk.
El PONENT:
Joaquim João Júdice is a Retired Full Professor of the Department of Mathematics of
University of Coimbra, since November 2011. He is also a current member of the Instituto de
Telecomunicações of Portugal. He is an Associate Editor of 13 international research journals in
the areas of Mathematics and Operational Research. He was referee of several papers submitted
to 38 different international journals. He was in the Organizing Committee of 22 national and
international conferences and in the Scientific Committee of 58 conferences. He was Director of
the Department of Mathematics of the University of Coimbra for two years, President of Centro
Internacional de Matem´atica (CIM) for four years, President of the Portuguese Operational Society
(APDIO) for two years and Vice-President of the Portuguese Mathematical Society (SPM) for
two years. He received the 2012 EUROPT Fellow. He is author or co-author of 64 papers in
international research journals, 14 in national journals with referee and 23 in edited volumes with
referee and of 19 books and lecture notes.
Clica aqui per més informació sobre Joaquim Júdice.




Inici
Routing problems with profits
CONVIDADA:
M. Grazia Speranza
IDIOMA: Anglès
LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)
DATA: Divendres, 16 de novembre de 2012. Hora: 12:30
RESUM:

While the literature is rich of papers that study routing problems where all customers to be served are given, the number of papers on routing problems with profits is much more limited. These problems model situations where a set of customers has to be selected, and have a number of interesting applications, in particular in the context of auctions in the transportation sector. The interest in these problems is growing. Work has been done in the case of one vehicle, while very little is available for the case of multiple vehicles.

In this talk I will overview the state of the art in routing problems with profits, introducing the basic problems of this class. I will then focus on some recently studied problems. I will present two capacitated node routing problems and one arc routing problem with profits. I will also present solution algorithms, both exact, namely a branch-and-price, and heuristic. While the exact algorithms can solve, for each of the problems, instances of small size, the heuristics obtain very good results in terms of solution quality, within a reasonable amount of time. The optimal solution on small instances allows us to evaluate, at least on those instances, the behaviour of the heuristics precisely.

 

LA PONENT:


M. Grazia Speranza és catedràtica del departament "Quantitative Methods" de la Universitat de Brescia, Itàlia.   Durant la seva carrea ha ocupat diferents càrrecs acadèmics o de gestió tant dins com fora de la seva universitat i, en l'actualitat, presideix EURO, l'associació de societats europees d'investigació operativa.
La seva recerca està orientada a l'optimització combinatòria. En aquest àmbit, ha treballat en la resolució tant exacta com heurística de diferents problems, especialment en problemes de disseny de rutes de vehicles i en problemes de finances i la seva recerca ha donat lloc a més d'un centenar de publicacions en revistes de prestigi.
Per més informació, clica aqui.




Inici


Recent advances in phase II/III clinical trials: Study Design and Methods of Analysis
CONVIDAT: Sue Todd

IDIOMA: Anglès

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: Dijous, 15 de novembre de 2012. Hora: 12:00

RESUM: There has been much recent interest in the development of clinical trial methods that enable adaptation of the design during the course of the trial whilst maintaining statistical and scientific integrity through the control of type I error rate (see Friede and Stallard [Biometrical Journal (2008) 50 767-781] for a comparison of various methods). A particularly important type of adaptation for the setting of clinical trials is the ability to stop the trial early or to select treatments during the trial. Using this approach, the simplest form of design is conducted in two stages. In the first stage, patients are randomised between a placebo and a number of experimental treatments. At the end of this stage, an interim analysis of the data observed so far, possibly on a short-term or surrogate endpoint, will be conducted. As a result of this, the trial could be stopped early, or one or more of the experimental treatments may be dropped from the trial. The trial then continues with further patients randomised between the placebo and the remaining experimental treatments. At the end of the trial, these experimental treatments are then compared with the placebo using the data from all patients randomised. The final statistical inference is conducted so as to maintain the overall type I error rate allowing for the treatment selection and interim analyses.
This talk describes the methodology which has been developed for these types of trials, considers the use of simulation studies as a powerful tool for exploring the properties of different treatment selection designs and presents an example of the use of an adaptive phase II/III clinical trial design in practice.



Inici


Un algoritmo de branch-and-cut para un problema de rutas con un camión, un remolque y puntos de recarga

(STTRPSD)

CONVINDAT: Enrique Benavent
IDIOMA: Castellà
LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)
DATA: divendres 19 d'octubre HORA: 12:30

RESUM:

En el STTRPSD se dispone de un camión con un remolque desmontable que, partiendo de un depósito, tiene que servir la demanda de un conjunto de clientes que son accesibles sólo con el camión. Así, antes de servir a los clientes, es necesario desmontar el remolque en lugares de aparcamiento apropiados (puntos de recarga) y transferir productos entre el camión y el remolque. La ruta de servicio se puede descomponer entonces en: una ruta que sigue el camión con el remolque que pasa por los puntos de recarga elegidos y un conjunto de rutas, para el camión sólo, que salen de los puntos de recarga y que realizan el servicio a los clientes. Este problema tiene aplicaciones por ejemplo en la recogida de leche en granjas que no son accesibles por vehículos de gran tamaño.

En este trabajo se presenta una formulación del problema como un problema lineal entero. Esta formulación es reforzada con varias familias de desigualdades válidas para las que se ha diseñado diversos algoritmos de separación. Todo ello es la base de un algoritmo de branch-and-cut que permite resolver el STTRPSD para instancias de tamaño medio (hasta 100 nodos) y proporciona cotas inferiores ajustadas en instancias de mayor tamaño.

Éste es un trabajo conjunto con J.G. Villegas, J.M. Belenguer, A. Martínez, C. Prins y C. Prodhon

 

CONVIDAT: El profesor Benavent es Doctor en Matemáticas por la universitat de València con un trabajo sobre el Problema de Asignación Cuadrática en 1982 y Profesor Titular en la misma universidad desde el año 1984. Su docencia e investigación se ha centrado en el área de la Investigación Operativa, principalmente en temas de Optimización Discreta. Ha trabajado sobre todo en problemas de Rutas de Vehículos con publicaciones en revistas internacionales del área. Sus líneas de investigación se centran en: Optimización Discreta, Programación Entera, Teoría de Grafos, problemas de Rutas de Vehículos y Localización.



Inici



Generating solutions to practical decisión problems by dynamic programming

CONVINDAT: Eligius M.T. Hendrix
IDIOMA: Anglès
LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)
DATA: divendres 21 de setembre HORA: 12:30

RESUM:

Dynamic programming focuses on the question when, in which situation to take which decision. In our experience it is not straightforward for students and researchers to grasp its concepts and to come to workable implementations. In this presentation we discuss its use for answering several practical questions from engineering and economics.

• How long should queues before traffic lights be in order to switch colour?

• When to release water in reservoir or lake management?

• When to refill the beer in your fridge?

• Is this similar for salad (perishable inventory products)?

• Which actions to take consecutively in an agro-logistics supply chain?

• How fast should the European Union revise fish quota?

• Does price variation influence deforestation in Latin America?

• Will pollution of several industries in Australia lead to closure of some?

We go through the steps and challenges to answer these questions via dynamic programming

 

CONVIDAT: (Clica aqui per més informació)

Eligius M.Th. Hendrix és un investigador en mètodos d'optimizació amb un contracte Ramon y Cajal a la Universidad de Málaga, i també està associat a la Universidad de Wageningen, Holanda. En la seva experiència docent de 25 anys ha supervisat prop de 100 tesis de màster i vàries tesis de doctorat. És autor de nombrosos articles en revistes internacionals i de contribucions en diversos llibres i articles nacionals. Les seves contribucions van des de qüestions sobre com implementar algoritmes en plataformes modernes de manera eficient per a problemes teòrics de teoria de jocs, fins a ciències agrícoles i mediambientals, logística agrària, disseny d'experiments, anàlisi de riscos, gestió d'aigües, industria del farratge, et. La majoria de les qüestions fonamentals de recerca estan lligades amb optimització global, optimització robusta, formació de coalicions i problemes de localització; àmbits en què també organitza tallers i escriu llibres.


Inici


Functional Analysis of Cardiovascular Data
CONVIDAT: Emilia Bagiella

IDIOMA: Anglès

LLOC: FME, Edifici U, Aula: 102, Campus Sud, UPC, Pau Gargallo, 5, 02028 Barcelona (veure mapa)

DATA: Dijous, 5 de juliol de 2012. Hora: 13:00

RESUM: Heart rate continuously fluctuates over time either spontaneously or in response to external stimuli. These continuous oscillations produce variability of the cardiovascular parameters. The assessment as well as the interpretation of this variability has captured the interest of many researchers. Fourier-based approaches to analysis of variability of RR intervals (RRV) typically compute power in a given frequency band (e.g. 0.01-0.07 Hz) by aggregating the power at each constituent frequency within that band. In this way a large amount of data is reduced to a single discrete measure. The approach over-simplifies the complex structure of the data and therefore results in loss of information and, correspondingly, statistical power to detect effects and associations. We propose to represent RRV as continuous functions of the frequency domain rather than discrete points and to explore the use functional data analysis to analyze these curves. We use a wavelets transform to reduce the dimensionality of the data and a wavelet-based LASSO approach to fit linear models with functional predictors.


Inici

Spatio-temporal modelling of wildfires in Catalonia through log Gaussian Cox processes

CONVIDAT: Marc Saez

IDIOMA: Català

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: Divendres, 29 de juny de 2012. Hora: 12:30

RESUM: Forest fires management is not only an emergency task, the preventive task could be even more important, being better avoid risk of a forest fire ignition before it start or minimize its hazard, that latter try to extinct it. If we associate wildfires with their spatial coordinates, along with other variables, it is possible to identify them by means of a spatio-temporal stochastic process. Spatio-temporal clustering of wildfires could indicate the presence of risk factors. In fact, what is usually of interest is to assess their dependence on covariates. Two were the objectives in this paper. Firstly, to evaluate how the extent of clustering in wildfires differs across marks. Secondly, to analyze the influence of covariates on trends in the intensity of wildfire locations. We analyzed the spatio-temporal patterns produced by wildfire incidences in Catalonia, located in the north-east of the Iberian Peninsula. The total number of forest fires (over 0.5 wooded hectareas burned or more than 1 hectarea in total burned) recorded in the studied area, during the period 1994-2008, was 2449 (forest sur. In addition to the locations of the fire centroids, several marks and spatial covariates were considered. We specified spatio-temporal log-Gaussian Cox process models. Models were estimated using Bayesian inference for Gaussian Markov Random Field (GMRF) through the Integrated Nested Laplace Approximation (INLA) algorithm. Results allow us to quantify and assess possible spatial relationships between the distribution of risk of ignition and possible explanatory factors. We believe the methods shown in the paper may contribute to the prevention and management of wildfires, which are not random in space or time.

El ponent:
clicar aquí per més informació sobre les linies d'investigació d'en Marc Saez



Inici

Data Analytics of Narrative: Pattern Recognition in Text, and Text Synthesis, Supported by the Correspondence Analysis Platform.

CONVIDAT: Fionn Murtagh

IDIOMA: Anglès

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: Divendres, 25 de maig de 2012. Hora: 12:30

RESUM: I present a short overview of the Correspondence Analysis platform for analysis of text. This includes: metric, factors, contributions. Using the Euclidean factor space, sequence-constrained agglomerative hierarchical clustering is described. I then look at applications of this data analysis platform. First there is the Casablanca movie, where emotion is tracked and displayed. Next statistically-based style analysis is briefly discussed. Such work can be said to be analysis of narrative. I next look at synthesis of narrative. Collective authoring is at issue here. Finally I take an example from online digital content. Using a set of blog postings, I discuss influence, and change and orientation of the narrative. My aim in this work is as follows. As opposed to the current standard model where authority and influence are gauged by numbers of hits (or accesses, or citations) - let me call this the "likeability model" - I instead pursue a semantic analysis where influence is gauged by role in its effect on the subsequent activities, e.g. blog discussion. The tools for quantifying such semantic rating are provided by moments of inertia and related measures that are the basis of geometric data analysis.

References:

B. Le Roux and H. Rouanet, Geometric Data Analysis: From Correspondence Analysis to Structured Data Analysis, Springer, 2004.
B. Le Roux and H. Rouanet, Multiple Correspondence Analysis, (Series: Quantitative Applications in the Social Sciences, Volume 163), SAGE Publications, 2010.
F. Murtagh, Correspondence Analysis and Data Coding with R and Java, Chapman & Hall/CRC, 2005.
F. Murtagh, A. Ganz and S. McKie, The structure of narrative: the case of film scripts, Pattern Recognition, 42, 302-312, 2009. See discussion: Z. Merali, Here's looking at you, kid, Nature, 453, p. 708, 4 June 2008.
F. Murtagh, A. Ganz, S. McKie, J. Mothe and K. Englmeier, Text sequence visualization using planar maps, hierarchical clustering and tag clouds: the case of filmscripts, Information Visualization, 9, 253-262, 2010.
F. Murtagh, A. Ganz and J. Reddington, New methods of analysis of narrative and semantics in support of interactivity, Entertainment Computing, 2, 115-121, 2011.

El ponent: clicar aquí per més informació sobre les linies d'investigació d'en Fionn Murtagh


Inici
Locating Lottery Retail Stores in Spain
CONVIDAT: Daniel Serra

IDIOMA: Castellà.

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: divendres, 18 de maig de 2012.  Hora: 12:30

RESUM: A Lottery ticket is a very homogeneous good, with no price or product differentiation, and therefore the location of lottery outlets is a key factor in the success of the business. In Spain there are almost 10.000 retail stores that sell lottery. In this study we present a model to locate new stores so as to maximize customers' capture, but with the constraint of trying to minimize the impact on existing customers. Results on its implementation in Spain are presented. 

 

El ponent:

Daniel Serra graduated in 1984 in Economics from the Autonomous University of Barcelona, and obtained a master in systems analysis and his PhD in the Whiting School of Engineering at Johns Hopkins University in 1989. He is actually professor of management at the Universitat Pompeu Fabra (UPF). His fields of specialization are logistics and quantitative methods in management. He has more than 30 publications in international journals, such as European Journal of O.R., Computers and O.R., Journal of the Operational Research Society, Network and Spatial Economics, Journal of Regional Science, Geographical Analysis, Papers in Regional Science, among others.  He belongs to the editorial board of Geographical Analysis, International Journal of Regional Science, Supply Chain Practice, and International Journal of Operations Research and Information Systems. He is actually the director of the Business Analytics Research Group at UPF and vicerector for Economics, Information Systems and Institutional Relations.


clicar aquí per més informació sobre Daniel Serra


Statistical Aspects of Google Flu Trends


CONVIDAT:
Samantha Cook

IDIOMA: Anglès

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: Divendres, de 2012. Hora: 12:30
RESUM: Google Flu Trends uses anonymized, aggregated internet search activity to provide near-real time estimates of influenza activity. Google Flu Trends estimates have shown a strong correlation with official influenza surveillance data in the United States and around the world. We present the Flu Trends methodology; evaluate its performance, including the H1N1 outbreak of 2009; and suggest directions for future developments.

 


Inici
Detection and investigation of carcinogenic risk exposures using the observed temporal pattern of the associated events.

CONVIDAT: Rina Chen

IDIOMA: Anglès

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: Divendres, 27 d'abril de 2012.  Hora: 12:30
RESUM: Health professionals are often required to analyze data when the incidence of a cancer disease is suspected to be elevated. Because of several reasons (ad hoc analyses, small number of events spread over time, etc.) both types of statistical errors are large in such analyses.  Thus, whether or not the observed/expected number of events is significant, further steps should be applied to the data.  A significant result may call for an epidemiologic investigation that enables distinction between false and true alarms; the possibility that a non-significant result is a missed true alarm should be considered. The q interval and the CUSCORE test may alleviate these problematic issues. The important feature of interval and the test is their sensitivity to temporal clustering of the events.  Clearly detection of a cluster within the dates of diagnoses points to a true alarm.

The q interval is derived from a standardized time interval.  It follows the uniform distribution over values in the range 0-1.  A unit of the standardized time interval is defined as the expected calendar time between each two consecutive events.  As such it is a standard unit that does not depend on the incidence rate nor on changes in the population at risk that usually occur over time.  The CUSCORE test is a sequential type procedure based on that standard interval.  The test is easily applicable to any distributed variable. Actually it has been applied to a gamma-distributed variable.  The q interval and the CUSCORE test will be presented and demonstrated using real data sets.



Inici
Análisis forense de datos electorales.
CONVIDAT: Raúl Jiménez

IDIOMA: Castellà.

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: divendres, 20 de Març de 2012.  Hora: 12:30
RESUM: Conversaremos sobre el problema general de análisis forense de elecciones pero centrándonos en el enfoque estadístico para explorar, después del evento electoral, si existen irregularidades y sesgo en el conteo de votos. Revisaremos la reciente literatura sobre esta controvertida área de investigación y estudiaremos algunos casos emblemáticos, haciendo énfasis en el Referéndum Revocatorio Venezolano del 2004, que muy recientemente ha sido reseñado de nuevo en una sección especial de Statistical Science.

 

El ponent: clicar aquí per més informació sobre les linies d'investigació d'en Raúl Jiménez.


Inici
Stochastic programming with mixed-integer variables.
CONVIDAT: Maarten van der Vlerk

IDIOMA: Anglès.

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: dimecres, 18 de Març de 2012.  Hora: 12:30

RESUM: La programació estocàstica estudia models i mètodes de resolució per donar suport a la presa de decisions sota incertesa, amb la intenció de tenir en compte de manera explícita una característica important de molts problemes del món real. En lloc d'assumir que els paràmetres futurs són coneguts, s'assumeix que només es disposa d'informació probabilística sobre aquestos a l'hora de prendre les decisions. Farem una revisió d'alguns dels principals models estudiats en programació estocàstica. Entre aquests, ténen un interès particular aquells que involucren variables senceres (binàries), com és el cas en moltes situacions pràctiques. Per exemple, la decisió sobre la localització de centres de distribució es prodria plantejar com un programa estocàstic amb variables senceres. Malhauradament, la resolució d'aquest tipus de problemes suposa un repte important, donat que combinen les dificultats pròpies de la programació estocàstica i la programació sencera. En aqeust seminari es discutiran alguns dels desenvolupaments més recents en aquesta activa àrea de recerca.

 

El ponent: Maarten H. van der Vlerk és catedràtic d'Optimització Estocàstica a la Universitat de Groningen (Holanda). És el director del grau Bsc. en Econometria i Investigació Operativa i del màster en Econometria, Investigació Operativa i Ciències Actuarials d'aquesta universitat. La seva principal area de recerca és la programació estocàstica sencera, àrea en què ha treballat molt activament. Entre altres coses, ha compilat i mantingut una bibliografia sobre programació estocàstica que actualment conté més de 4000 referències, ha estat secretari i posteriorment director de la societat COSP i, juntament amb David Morton, va fundar la web dedicada a programació estocàstica stoprog.org. A més, ha publicat més de 40 treballs en les revistes més rellevants de l'àrea.
clicar aquí per més informació sobre Maarten van der Vlerk.

 


Inici
Spatial prediction of nonstationary functional processes
CONVIDAT: Ramón Giraldo

IDIOMA: Castellà.

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: dimarts, 17 de març de 2012.  Hora: 12:30
RESUM: Two methodologies are proposed for carrying out spatial prediction of curves when the stationarity assumption is not valid because the mean function is not constant through the region of interest. In the first approach the curves are detrended by using a functional regression model with functional response and a geostatistical analysis is done with the residuals. We call this procedure residual kriging for functional data. A second approach is based on an extension of universal kriging to the functional context.

El ponent: clicar aquí per més informació sobre les linies d'investigació d'en Ramon Giraldo.

 


Inici
Global optimization strategies for detailed kinetic models in systems biology
CONVIDAT: Albert Sorribas

IDIOMA: Català.

LLOC: Edifici C5, Aula C5016, Campus Nord, UPC (veure mapa)

DATA: dijous, 15 de març de 2012.  Hora: 12:30

RESUM: Design of newly engineered microbial strains for biotechnological purposes would greatly benefit from the development of realistic mathematical models for the processes one wants to optimize. With the development and application of appropriate optimization techniques, one could identify the modifications that need to be made to the organism in order to achieve the desired biotechnological goal. As appropriate models to perform such an analysis are necessarily non-linear and typically non-convex, finding their global optimum is a challenging task. Canonical modeling techniques, such as Generalized Mass Action (GMA) models based on the power-law formalism, offer a possible solution to this problem because they have a mathematical structure that enables the development of specific algorithms for global optimization.

Based on the GMA canonical representation, we have developed in previous works a highly efficient optimization algorithm and a set of related strategies for understanding the evolution of adaptive responses in cellular metabolism. Here, we explore the possibility of recasting kinetic non-linear models into an equivalent GMA model, so that global optimization on the recast GMA model can be performed. With this technique, optimization is greatly facilitated and the results are transposable to the original non-linear problem. This procedure is straightforward for a particular class of non-linear models known as Saturable and Cooperative (SC) models that extend the power-law formalism.

Our results show that recasting non-linear kinetic models into GMA models is indeed an appropriate strategy that helps overcoming some of the numerical difficulties that arise during the global optimization task.

References

- Pozo C, Marin-Sanguino A, Guillén-Gosalbez G, Jimenez L, Alves R, Sorribas A. Steady-state global optimization of non-linear dynamic models through recasting into power-law canonical models. (2011) BMC Systems Biology 5:137.

- Sorribas A, Pozo C, Vilaprinyo E, Guillén-Gosálbez G, Jimenez L, Alves R. Optimization and evolution in metabolic pathways: global optimization techniques in Generalized Mass Action models. Journal of Biotechnology 149(3):141-153

- Guillén-Gosálbez-G, Sorribas, A. Identifying quantitative operation principles in metabolic pathways: a systematic method for searching feasible enzyme activity patterns leading to cellular adaptive responses. BMC Bioinformatics 2009, 10:386.


El ponent:
clicar aquí per més informació sobre les linies d'investigació d'Albert Sorribas.



Inici

El problema del cartero chino con beneficio máximo
CONVIDAT: Ángel Corberán

IDIOMA: Espanyol.

LLOC: Edifici C5, Aula: C5016, Campus Nord, UPC (veure mapa)

DATA: Divendres, 27 de gener de 2012, Hora: 12:30

RESUM: El Problema del Cartero Chino con Beneficio Máximo (MBCPP) es un problema NP-difícil que considera varios beneficios asociados con cada arista de un grafo, uno por cada vez que la arista es atravesada realizando el servicio. El objetivo es encontrar un tour con beneficio máximo. El MBCPP tiene su aplicación en la resolución de problemas relacionados con la retirada de nieve de calles y carreteras. Proponemos una formulación como PLE del MBCPP definido sobre un grafo no dirigido y, basándonos en la descripción parcial que hemos obtenido de su poliedro asociado, proponemos un método de ramificación y acotación para su resolución. Presentaremos resultados computacionales en instancias con hasta 1000 vértices y 3000 aristas.
El ponent:

Ángel Corberan és professor Catedràtic d'Universitat del departament d'Estadística i Investigació Operativa de la Universitat de València. La seva recerca s'inscriu en l'àrea d'optimització combinatòria/programació discreta, fonamentalment en problemes de distribució, transport i disseny de rutes de vehicles. És autor d'un bon nombre d'articles publicats en revistese de prestigi relacionats amb l'estudi dels problemes esmentats, i el disseny i implementació de mètodes aproximats i exactes per a la seva resolució. És editor de Computational Optimization and Applications, Computers & Operations Research i TOP
clicar aquí per més informació sobre Ángel Corberán.

 


Inici